Professor Jered Borup started his lecture for future teachers on cutting-edge artificial intelligence by showing a nearly 50-year-old commercial advertising a culinary breakthrough: The microwave.
The 1978 spot featured a pixie cut blonde marveling at an appliance that could make Thanksgiving dinner in minutes.
Then came the 2026 reality check: “Who here has cooked a Thanksgiving turkey in the microwave?” Borup asked the roomful of George Mason University undergraduate elementary education majors.
There were some giggles. But not a single hand shot up.
Microwaves were once billed as a do-it-all, super-efficient technology, just as AI is now, Borup said. These days, no cook would rely just on their microwave to prepare a holiday feast.
But they’re a go-to for reheating a slice of pizza.
The message behind the metaphor: AI can be helpful for teachers in limited circumstances, as a brainstorming partner, and may save educators time on certain rote tasks —the instructional equivalent of nuking leftovers. But it will never replace the conventional oven of teacher judgment.
Educators’ goal in using AI is “to make instruction more efficient, engaging, effective,” Borup told his students. “But with any tool, if you use it poorly, you can actually decrease the efficiency, the engagement, and effectiveness” of instruction.
Unlike Thanksgiving turkey, there’s no trove of tried-and-true recipes for the challenge Borup and his colleagues in GMU’s College of Education and Human Development have taken on: Helping current and future educators—including undergraduates who’ve never been in front of a K-12 classroom—navigate the pitfalls and potential of a rapidly developing, controversial technology for teaching and learning.
Research on AI’s impact on K-12 instruction is in the early stages. It’s not clear what K-12 districts want incoming teachers to know about AI. Some school districts have embraced AI. Others ban it.
But Ingrid Guerra-López, the college’s dean, has her own vision: She’d like aspiring teachers to leave GMU with the AI knowledge and skills to be able to explain the tech’s power and limitations to their future students, including its tendency towards biased or inaccurate information.
And she’d like prospective teachers at GMU to understand how to use the tech to bolster student learning—for instance, in analyzing assessment data to pinpoint next steps for instruction.
“The conversation about AI and education is more about education than about AI,” Guerra-López said. “AI is a means to an end. It is a tool, but the focus is always strengthening teaching and learning.”
GMU partners with local districts to think through AI use
Helping students navigate this increasingly ubiquitous technology is a priority for GMU as an institution, not just for its teacher education program, Guerra-López said.
All undergraduates—whether they’re majoring in education, kinesiology, or art history—must choose a technology course as a general education requirement. Some options include an AI focus.
GMU’s college of education and human development also has a host of AI initiatives or teaching, learning, or research programs that include an AI twist.
For instance, the school is working to develop a “resource hub” for local districts featuring information on AI tools. And GMU is aiming to offer a 15-credit graduate certificate in learning technologies, with a concentration in AI teaching and learning.
The college also has a research and practice partnership—the Educational Research Alliance of Northern Virginia or ERA-NOVA for short—aimed at helping some 20 neighboring districts think through common challenges. Recently, much of that work has centered on AI implementation.
For instance, at an October session, GMU experts explained a new state law that requires districts to develop AI guidelines, based on a forthcoming model policy from the state education agency. Then, leaders from a couple districts shared tips for addressing families’ AI concerns.
Students aren’t required to take an AI course, but many do
Still, as in most teacher-prep programs nationally, there’s no hard-and-fast rule that preservice teachers get background in AI before they graduate from GMU.
Adding a required AI class—or a mandated course on any subject for that matter—would be a heavy lift, in part because educator prep programs must adhere to state and national accreditation standards, Guerra-López said.
Teacher-prep programs “have very little flexibility on how much [they] can veer off the [list of] courses you actually have to have,” she said.
But many faculty members have chosen to add AI content to long-required courses, Guerra-López said.
For example, Borup and at least two other professors who lead technology courses for pre-service elementary school teachers put some focus on AI, though their lectures and assignments aren’t identical.
GMU—like many university-based teacher-prep programs—isn’t in the business of forcing faculty to use AI or to teach future educators about the tech, Guerra-López said.
In higher education “curriculum is very much owned by the faculty,” Guerra-López said. Professors have “creative freedom to design and deliver a course in the way that they think is appropriate from their point of view, as a scholar.”
That means instruction on AI “varies from course to course,” Guerra-López said.
Some professors prefer to steer clear, citing ethical issues with the technology, including allegations that companies violated intellectual copyright laws to develop large language models, Guerra-López said.
Others find AI plain overwhelming.
Learning to use AI “means that they’re going to have to upskill, and not everybody feels comfortable with technology,” Guerra-López said. “And then you have other folks that are just so eager to adopt and try new things. Either way, they need support.”
More than three-quarters of teacher educators aren’t receiving sustained professional development on AI tools, according to an EdWeek Research Center survey of 268 teacher educators conducted in August and September.
GMU’s faculty have the opportunity to be an exception.
Two education professors—Borup, who specializes in instructional design, and Jennifer Suh, a professor of math education—teamed up with a computer science colleague and a professor specializing in AI ethics and policy to offer a series of webinars to any interested faculty in the education school.
About 30 instructors have shown up regularly while others drop in. Participating professors got an overview on how the tech works and the chance to play around with AI tools.
One experiment involved creating instructional materials in their own area of expertise, then discussing what the tech did well and what it missed, Suh said.
In the spring, the group will launch a “lesson study” on AI, including sharing ideas for teaching future educators about the tech, and inviting colleagues to drop into their classes to watch them in action.
Professors want their colleagues’ thoughts on “this new phenomenon that nobody really has figured out because every day it’s changing,” Suh said.
Aspiring teachers question the value of AI for critical elements of instruction
Borup’s class for undergraduates also asks students to train a critical eye on AI-created instructional resources.
As an out-of-class assignment, Borup’s directed students to experiment with several popular AI platforms designed for K-12 education to develop lesson plans, assessments, classroom activities, and other learning material.
Self-described AI skeptic Evie Madison Perks, a sophomore and aspiring special education teacher, didn’t expect to enjoy the process.
But she found it “fascinating” to work with AI to create a quiz for 4th graders about a topic she knows well: Country Singer Kacey Musgraves.
“This is cool,” Perks said on Zoom, as she completed the assignment while a reporter watched.
Still, Perks doesn’t plan to use AI to create instructional resources for her undergraduate courses unless she’s directed to.
For one thing, she worries about AI’s environmental impact.
She also doubts she’ll use AI much as a teacher.
“I’ve always enjoyed putting in hard work,” Perks said. “I just don’t see myself using it to do critical [teaching tasks] of my job.”
In a class discussion of the assignment the following day, Perk’s classmates shared similar qualms.
One worried that the open-ended questions an AI tool wrote for a quiz for kindergartners were too advanced. Another found a feature aimed at leveling text for different reading abilities to be a bust.
And a future special education teacher was taken aback when she saw a tool could help craft an Individualized Education Program. Putting that information into AI could violate students’ privacy, she said. (Borup told her she was right to be concerned.)
Stacy Reyes, an aspiring elementary school teacher, said she’s skeptical AI would save her much time, given how much she’d need to factcheck and tweak its output to fit her teaching style.
And she doesn’t feel like she has the know-how, at this early point in her development as a teacher, to gauge how well a lesson or quiz that looks great on an AI platform would go over with real students.
“I would have to be in the classroom to see if it would be successful,” Reyes said in an interview during a class break.
Borup expects that his students will be more sophisticated at critiquing AI-created instructional materials once they get more hands-on teaching experience.
For now, Borup offered a tip: Think of AI as a human assistant. If you just told that person: “get me lunch,” they could return with almost anything. But if you say: “go to Panera and get me a number 2,” they’ll be more likely to bring back something you’ll enjoy eating.
“The more specificity you can include in your request, the fewer decisions [AI] has to make, and the more likely you’re going to get what you want,” Borup explained.
Next, Borup helped his students explore some of AI’s inner-workings.
Students spent a few moments in class using an online tool to create and train simple machine learning models that could distinguish between a happy and sad expression or thumbs and thumbs down, or one person’s face from another’s.
The training didn’t always lead to correct answers—a sign of the technology’s fallibility.
And he talked about the dangers of children possibly creating relationships with AI tools. He explained that the tech is programmed to keep users engaged. That can lead to sycophantic, even dangerous responses from the AI.
To demonstrate, Borup pulled up the audio version of a large language model in class and told it he was lonely.
“I can be your friend,” the model said, spurring uncomfortable laughter in the classroom.
The finale: Borup shared a conversation in which a chatbot endorsed his plan to raid his savings account to make a movie version of a boring lecture.
Students will continue to learn about AI in teaching and learning throughout the course, going over concepts like “vibe coding” and using AI prompts to create classroom resources and lesson plans that embrace the principles of Universal Design for Learning, Borup said.
District leaders see shared responsibility in building teachers’ AI literacy
Dustin Wright, the superintendent of the City of Fairfax school district, a short drive from George Mason’s campus, said the university’s collaborations with local school systems—including through the ERA-NOVA research partnership—has helped university faculty understand district leaders’ perspectives on what educators need to know to move towards “responsible innovation” with AI.
Wright, who holds multiple degrees from George Mason, served as an adjunct professor there, and has hired some of its graduates, wants teachers—current and future—to consider this question: “How can we incorporate these [AI] tools while also preserving critical thinking, creativity, rigor, and just good judgment, whether it’s the judgment of the adults or the judgment of our students?”
He’s glad that conversation can begin at the pre-service level. But, Wright added, that no district leader would argue that it is “completely the responsibility of the teacher-prep programs to provide all the training for our educators” on AI or anything else.
“That’s the start of it,” he said. “Then we pick it up, once they start working for us.”